Treatment of adult ALL patients with third-generation CD19-directed CAR T cells: results of a pivotal trial

Abstract

Background: Third-generation chimeric antigen receptor (CAR)-engineered T cells (CARTs) might improve clinical outcome of patients with B cell malignancies. This is the first report on a third-generation CART dose-escalating, phase-1/2 investigator-initiated trial treating adult patients with refractory and/or relapsed (r/r) acute lymphoblastic leukemia (ALL).

Methods: Thirteen patients were treated with escalating doses of CD19-directed CARTs between 1 × 10⁶ and 50 × 10⁶ CARTs/m². Leukapheresis, manufacturing and administration of CARTs were performed in-house.

Results: For all patients, CART manufacturing was feasible. None of the patients developed any grade of Immune effector cell-associated neurotoxicity syndrome (ICANS) or a higher-grade (≥ grade III) catokine release syndrome (CRS). CART expansion and long-term CART persistence were evident in the peripheral blood (PB) of evaluable patients. At end of study on day 90 after CARTs, ten patients were evaluable for response: Eight patients (80%) achieved a complete remission (CR), including five patients (50%) with minimal residual disease (MRD)-negative CR. Response and outcome were associated with the administered CART dose. At 1-year follow-up, median overall survival was not reached and progression-free survival (PFS) was 38%. Median PFS was reached on day 120. Lack of CD39-expression on memory-like T cells was more frequent in CART products of responders when compared to CART products of non-responders. After CART administration, higher CD8 + and γδ-T cell frequencies, a physiological pattern of immune cells and lower monocyte counts in the PB were associated with response.

Conclusion: In conclusion, third-generation CARTs were associated with promising clinical efficacy and remarkably low procedure-specific toxicity, thereby opening new therapeutic perspectives for patients with r/r ALL. Trial registration This trial was registered at www.

Clinicaltrials: gov as NCT03676504.

Keywords: Acute lymphoblastic leukemia (ALL); CART-associated toxicities; CD39; Cytokine release syndrome (CRS); Cytopenia; Immune effector cell-associated neurotoxicity syndrome (ICANS); Investigator-initiated trial (IIT); Third-generation chimeric antigen receptor (CAR) T cells.

Fig. 1

HD-CAR-1 study profile. Fifteen patients with relapsed and/or refractory (r/r) acute lymphoblastic leukemia (ALL) after at least two prior therapy lines were screened and enrolled into HD-CAR-1. For all patients, leukapheresis and manufacturing of CARTs were feasible. Two patients did not receive the HD-CAR-1 CART product due to progressive disease (PD). Thirteen patients were treated with CARTs, with three patients receiving 1 × 10⁶ (dose level (DL) 1), three patients 5 × 10⁶ (DL2), four patients 20 × 10⁶ (DL3) and three patients 5 × 10⁷ (DL4) CARTs/m². Ten patients reached end of study (EOS) on day 90 after CART administration. Three patients died due to progressive disease (n = 2) or due to septic organ failure (n = 1) prior to EOS

Fig. 2

Hematologic toxicity of HD-CAR-1 treatment. A Cytopenia and B cell aplasia. On day 0, i.e., after lymphodepletion (LD) and before CART administration, 69% (n = 9) of patients were neutropenic (46% grade IV neutropenia), anemic (8% grade III anemia) and thrombocytopenic (31% grade III thrombocytopenia). One month after HD-CAR-1 treatment, 8% (n = 1, UPN#4) patients displayed grade IV neutropenia and 17% (n = 2; UPN#4 and UPN#11) grade IV thrombocytopenia. On end-of-study (EOS) at day 90, two patients showed persistent grade III neutropenia (UPN#2, UPN #4) and one patient grade III thrombocytopenia (UPN#4) despite treatment with granulocyte-colony-stimulating factor (G-CSF) and a thrombopoietin-agonist, respectively. (UPN#4 had grade III neutropenia and thrombocytopenia already before receiving CARTs; also, UPN#2 was already neutropenic before CART treatment.) No higher-grade anemia was observed. Beyond day 28, no grade IV cytopenia was observed. As for B cell counts, 77% (n = 10) of patients displayed B cell aplasia already before receiving CARTs. At EOS, all evaluable patients (n = 9; UPN#9 not shown due to PD) had ongoing B cell aplasia (B cell count on day 0 and day 56 not assessed). B Absolute neutrophil count (ANC) of treated HD-CAR-1 patients (n = 13) within the first 18 days (top, small frame) and up to end of study on day 90 after CART treatment. Four patients (UPN #1, #3, #4, #11, #13) received G-CSF after CARTs. Median of ANCs is depicted in grey

Fig. 3

Efficacy of HD-CAR-1 treatment and patient outcome. A Overall survival (OS) and B progression-free survival (PFS) of treated patients. C OS and D PFS at end of study (EOS) on day 90 after HD-CAR-1 CART administration of HD-CAR-1 patients that achieved complete remission (CR; blue) vs. non-responders (red; partial remission (PR), stable disease (SD), progressive disease (PD). E Swimmer plot depicting the course of individual HD-CAR-1 patients. F OS and G PFS according to administered HD-CAR-1 CART dose (dose level (DL); DL1: 1 × 10⁶ CARTs/m² (n = 3), DL2 5 × 10⁶ CARTs/m² (n = 3), DL3 20 × 10⁶ CARTs/m² (n = 4), DL4: 5 × 10⁷ CARTs/m² (n = 3)). DL: dose level; CR: complete remission; and MRD (minimal residual disease). : CART therapy. : allogeneic stem cell transplantation. : antibody treatment. : chemotherapy. : progressive disease (PD), : partial remission (PR), : stable disease (SD), : MRD-positive complete remission (CR), : MRD-negative complete remission/metabolic CR (CR*), †: death

Fig. 4

Expansion of HD-CAR-1 CARTs. A Expansion of CARTs in the peripheral blood (PB) of individual HD-CAR-1 patients (n = 13) assessed by single-copy gene duplex quantitative PCR (SCG-DP-PCR) [19] after CART administration and up to end of study (EOS) at day 90. B Median expansion of CARTs according to administered CART dose levels (DL; DL1: 1 × 10⁶ CARTs/m², DL2: 5 × 10⁶ CARTs/m², DL3: 20 × 10⁶ CARTs/m², DL4: 5 × 10⁷ CARTs/m²). C Maximum CART copies (c_max) within 28 days after CART administration and clinical response at EOS (data of UPN#8 not shown due to progressive disease on day 23 after CARTs). Median (c_max) 22.350 CART/µg DNA PBMC. CR complete remission, d day, DL dose level, MRD minimal residual disease, PBMC peripheral blood mononuclear cell, PD progressive disease, UPN unique patient number

Fig. 5

Characterization of the cellular composition of CART products of HD-CAR-1 patients (n = 10). A CART infusion products were analyzed via high-parametric spectral flow cytometry, and data were analyzed (see methods). Uniform manifold approximation and projection (UMAP) visualization display a downsampled subset of cells from all ten CART products (bottom). After clustering, individual clusters were annotated based on surface marker expression [66] and highlighted by different colors. B CD8 + and CD4 + T cell subsets from the CART product of ten patient samples were extracted and clustered separately. A representative subset of cells from all ten CART products is displayed in the UMAP visualizations. Density plots in the two lower panels indicate the differential distribution of cells between non-responders (NR) and responders (R) within the CD8 + and CD4 + T cell compartment, respectively. C Boxplots indicating differential abundance of individual clusters from CD8 + (left) and CD4 + T cell (right) subsets from the CART product of responders and non-responders. Positive log₂ fold changes indicate higher levels in responders, whereas negative log₂ fold changes indicate that a specific population is more abundant in non-responders. D Principal component analysis (PCA) of CD8 + T cells within the CART product. Cell-type frequencies of cell clusters from each sample were used as input for the PCA. Blue circles represent samples from responders, and green circles represent samples from non-responders. The two larger circles indicate the midpoint of the respective group. Gray arrows indicate the variables. E Boxplots indicating the abundance of CD39- effector memory (EM)-like and CD39 + EM-like cells within the CD8 + T cell population of the CART products (left). A generalized linear mixed model (GLMM) was used to compute significance between non-responders and responders. Adjusted p values are shown. Boxplot of CD39 expression levels in non-responders and responders within the CD8 + T cell subset of the CART product is displayed (right). Significance was assessed by applying a linear mixed model (LMM). F PCA of CD4 + T cells within the CART product. Cell-type frequencies of cell clusters from each sample were used as input for the PCA. Blue circles represent samples from responders, and green circles represent samples from non-responders. The two larger circles indicate the midpoint of the respective group. Gray arrows indicate the variables. G Boxplots showing the abundance of CD39- EM-like and CD39 + EM-like cells within the CD4 + T cell population of the CART product (left). A GLMM was used to compute significance between non-responder and responder. Adjusted p values are shown. Boxplot of CD39 expression levels in non-responder and responder samples within the CD4 + T cell subset of the CAR product (right). Significance was assessed by applying a LMM. R responders, NR non-responders, CM central memory T cells, cDC conventional dendritic cells, EM effector memory T cells, NK natural killer

Fig. 6

Characterization of the cellular composition of PB samples (n = 10) of patients after HD-CAR-1 treatment (n = 10) and PB composition of healthy donors (n = 3). A PBMC samples obtained from patients after CART administration were analyzed via high-parametric spectral flow cytometry and data were analyzed (see methods). UMAP visualization (bottom) showing a downsampled subset of PBMCs from ten CART recipients and additionally three healthy donor samples. After clustering, individual clusters were annotated based on surface marker expression and highlighted by different colors. B Boxplots indicating differential abundance of individual cell populations from PBMC samples collected after CART administration, comparing abundances in responders and non-responders. Positive log₂ fold changes indicate that a respective population is more abundant in responders (R), whereas negative log₂ fold changes indicate that the population is more abundant in non-responders (NR). C Scatterplot displays the gating strategy to define CAR + cells. CD8 + and CD4 + T cells from the PBMC samples were extracted, and fluorescence intensity levels of CD8/CD4 expression were plotted against the fluorescence intensity of the CAR targeting antibody. CAR + cells were determined by setting a CD8 + /CD4 + T cell-specific cutoff for downstream analysis and visualization. D UMAP visualizations of downsampled subsets from separately clustered CD8 + and CD4 + T cells identified in A. Dimensionality reduction and clustering were performed excluding the expression information of the CAR targeting antibody, to prevent CAR + specific clusters. After clustering, individual clusters were annotated based on surface marker expression and highlighted by different colors. E Density plots illustrating the distribution of CAR + cells within the CD8 + T cell (top) and CD4 + T cell (bottom) UMAP embedding. CAR + cells were identified and gated as displayed in C and as described in the material and methods section. F CD4 + and CD8 + T cells from D were used and binned into CAR- and CAR + CD8 + or CD4 + T cells, respectively, as described above (Fig. 5C). Boxplots display differential abundance of different CAR + CD8 + T cells (top) or CAR + CD4 + T cell phenotypes (bottom) of responders and non-responders. Positive log₂ fold changes indicate that a respective population is more abundant in samples of responders, whereas negative log₂ fold changes indicate that the population is more abundant in samples of non-responders. R responders, NR non-responders, CM central memory T cells, cDC conventional dendritic cells, EM effector memory T cells, hi high, TCR T cell receptor, NK natural killer, NKT natural killer T cells, pDC plasmacytoid dendritic cells, SCM memory stem cell-like T cells